Results 54 comments of Liang Wang

This issue is related to this line of code: [https://github.com/yajiemiao/pdnn/blob/master/learning/sgd.py#L71](https://github.com/yajiemiao/pdnn/blob/master/learning/sgd.py#L71). `batch_size = 256`, which is much larger than size of training data `3`, leads to `train_sets.cur_frame_num / batch_size = 0`,...

`IOError: CRC check failed 0x3d667246 != 0x7f91b084L` It indicates your downloaded data file contains error, nothing to do with the code, you may delete plk.gz file and try again.

The diagonal elements of the resulting matrix correspond to positives. To be more specific, for the i-th row in `hr_vectors`, the positive is the i-th row in `tail-vectors`, and all...

我查了一下,是因为微博从http转向https了,导致url失效了。我已经在提交 https://github.com/intfloat/sina-weibo-crawler/commit/eb6975d94a0ebdff598b0adc9775414901f86871 修复了,你可以更新本项目代码,把`http://...` 换成 `https://...` 再试一下。

* 项目代码 git pull更新了么? * 你把cookie替换成你自己的了么?

那我想不出来还有什么问题了。。我这边本地测试是好的。

The in-batch negatives simply use samples from the same batch: https://github.com/intfloat/SimKGC/blob/97cc43e488f19ca5b0f6fbf60ffefd2ee56c0693/models.py#L91 The pre-batch negatives use cached negatives from previous batches: https://github.com/intfloat/SimKGC/blob/97cc43e488f19ca5b0f6fbf60ffefd2ee56c0693/models.py#L117-L133 The self-negatives use the head entities of each example:...

Not sure what you are referring to... Self-negative sampling does not add any new parameters.

> > Not sure what you are referring to... Self-negative sampling does not add any new parameters. > > May I know why you set -1e4 for masked_fill? Any number...

Sorry for the late response, I somehow missed this issue. **Q: Do we need hard negatives when finetuning for downstream tasks?** Yes, E5 fine-tuning needs hard negatives. For NLI, the...